2K output
The highest resolution EveryGen AI offers, sharp enough for product pages, ads and large screens.
Write a prompt or drop in a photo. EveryGen AI runs MiniMax H3 for you, renders up to 2K with sound and keeps every take in one chat.
MiniMax H3 is a video generation model from MiniMax. In EveryGen AI you use it through a chat: describe what you want, the agent turns it into a model prompt, chooses the settings and renders the clip, and you revise it with a reply.
MiniMax H3 makes clips from 5 to 15 seconds at 768P / 2K, from a text prompt, an opening and closing frame, or reference images. Every clip comes with a stereo soundtrack.
| Modes | Text to video, image to video (opening and closing frame) and reference to video |
|---|---|
| Clip length | 5–15 seconds |
| Resolution | 768P / 2K |
| Aspect ratios | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, or picked automatically |
| Image input | An opening frame plus an optional closing frame, JPG, PNG or WEBP up to 100 MB each |
| Sound | Generated together with the picture |
| Credits per 5 seconds | 80 credits at 768P; 100 credits at 2K |
| Output | MP4 video with audio, saved to the chat and your library |
The highest resolution EveryGen AI offers, sharp enough for product pages, ads and large screens.
A 5-second clip at 2K costs 100 credits, so the same budget buys more takes.
Each render includes a stereo soundtrack generated together with the picture.
Use an opening frame, plus an optional closing frame, and MiniMax H3 matches the video to the shape of your image.
Type what should happen, or attach an image. MiniMax H3 is already selected in the composer.
The agent writes the model prompt, sets length, ratio and resolution, and starts the render. You can leave the page; finished clips arrive in the chat and your library.
Reply with what to change, such as a slower camera, different light or a longer take, and the agent renders a new version next to the last one.
Pick Seedance 2.5 for longer takes and MiniMax H3 for 2K output at a lower cost. Both render sound, and you can switch between them in the same chat.
| Feature | MiniMax H3 | Seedance 2.5 |
|---|---|---|
| Clip length | 5–15 seconds | 4–30 seconds |
| Resolution | 768P / 2K | 480p / 720p |
| Aspect ratios | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
| 5-second clip | 100 credits at 2K | 400 credits at 720p |
| Longest clip | 300 credits | 2,400 credits |
| Sound | Generated with the picture | Generated with the picture |
Click a prompt to load it into the composer, then edit it before you send.
“Animate this portrait: she turns toward the camera and smiles, a light breeze moves her hair, soft window light, shallow depth of field.”
“Slow aerial push over misty mountain ridges at sunrise, golden light spilling into the valleys, birdsong and wind, 16:9.”
“Macro close-up of honey pouring onto a stack of pancakes, steam rising, warm kitchen light, slow motion, gentle sizzling sound.”
“Vertical 9:16 shot of a skateboarder rolling down a sunny beach boardwalk, the camera tracking alongside, upbeat street sounds.”
Yes. Use the composer on this page or in the chat, keep MiniMax H3 selected in the model menu and describe your video. You can switch to another model at any time in the same chat.
Between 5 and 15 seconds per clip. For longer takes, switch to Seedance 2.5 in the same chat.
768P / 2K. 2K is the default and the sharpest option in EveryGen AI; 768P costs fewer credits.
A 5-second clip at 2K costs 100 credits, and the cost grows with length. The composer shows the exact price before you send, and the pricing page lists every rate.
Yes. Attach an opening frame, and optionally a closing frame, as JPG, PNG or WEBP up to 100 MB each. MiniMax H3 animates the photo following your prompt.
No. EveryGen AI is an independent product that runs MiniMax H3 through a third-party API provider. MiniMax H3 is developed by MiniMax, and EveryGen AI is not affiliated with MiniMax.
MiniMax is a trademark of its owner. EveryGen AI is independent and is not affiliated with or endorsed by MiniMax.
Describe the shot or bring a photo, render it in 2K and revise it in chat until it is right.